SubsurfaceAI Inc’s Post

SeisWare software users can now import project data directly into SubsurfaceAI! We are pleased that this pipeline will help geoscientists be more productive and won't have to spend valuable time importing/exporting/reformatting data. Because who has time for that? #collaboration #software #geoscience

I am particularly proud of this release by SubsurfaceAI Inc. We put a lot of time in listening to our clients to speed up import and exporting as well as streamlining workflows. One of the most important new features is our brand new SeisWare International Inc. data pipeline. Being in both the #Canadian and #USA markets we feel this will greatly help our mutual customers. Also we keep improving our industry leading #interpAI product for #geoscientists. Some of the cool features are calculating AI probability volumes in 45 and 135 degrees away from inline when computing AI probability volume, as well as a new algorithm to track fault surfaces from fault probability volumes and export fault sticks. Please feel free to reach out to me to discuss this significant release. #machinelearning #seismic #geophysicist #geologist #AI

🌟 From Geomodeling Technology Corp. to SubsurfaceAI Inc.: A Journey of Innovation 🌟 Our journey began in the early 2000s when we developed specialist software like SBED and ReservoirStudio in collaboration with major oil companies to solve specific reservoir modeling challenges. These tools became industry benchmarks, empowering geoscientists to tackle complex subsurface problems with precision. 🏞️ Adapting to Change As the oil and gas industry evolves, so do we. Today, as SubsurfaceAI Inc., we’ve taken a bold step forward by unifying our expertise into the SubsurfaceAI platform — an integrated solution for subsurface interpretation, integration, and modeling, powered by cutting-edge AI and machine learning. 🚀 Empowering Subsurface Workflows Our focus is clear: increase the efficiency and accuracy of our end users. With innovations like seismic well ties, RGT and horizon cubes, and automated fault and horizon mapping, we’re enabling geoscientists to transition from traditional workflows to AI/ML-assisted processes that save time and reduce guesswork. 🤝 Built on Trust This transformation is possible thanks to the trust of our long-term users and early adopters of the SubsurfaceAI platform. Your success drives our commitment to deliver solutions tailored to today’s challenges. 🔗 Let’s Shape the Future We remain dedicated to supporting the industry with innovative tools and unparalleled service. Whether you're a long-time user or exploring AI/ML workflows for the first time, we invite you to join us in reshaping subsurface exploration and production. #AI #Seismicinterpretation#ReservoirModeling #Geoscience #Innovation #SubsurfaceAI Mark Tobin Renjun Wen Rongfeng Zhang Art Valdez Cindy Luo David Gray

The 3 Most Advanced Reservoir Modeling/Simulation Technologies(as of 2024) 1-tNav by Rock Flow Dynamics A relatively new software in the market(Introduced in 2005), offers great capabilities.Here is a short list of why people Love Tnav. **AI and ML Enabled. **Fast and Light Weight. **Fast Computation. **FDP Capabilities **Geological Modeling and Geostatistics **Deep Python Integration 2-Unified Ensemble Model by Halliburton Landmark Originally developed by Resoptima, UEM is a unique product in the market integrating Static and Dynamic models in Realtime with capability to tune and refine model simultaneously(at least this is how I understand it.).It's Called Ensemble due to many models bieng ran simultaneously (Simillar to RandomForest Algorithm in ML), and each model can correct the others till satisfactory uncertainty is achieved. Unique Features Found in This Solution: **Unique Concept and First in Market. **Uncertainity Based Modeling/Simulation. **ML Integrated and Enabled. **Wide Range of Dynamic Simulator Support 3-INTERSECT by SLB The Intersect high-resolution reservoir simulator is the answer to many of your reservoir challenges. By combining physics and performance in a fit-for-purpose reservoir simulator for your reservoir models. Here is Why INTERSECT is a lucrative option: **AI and ML Enabled. **Python Supported. **GPU Simulations Enabled. **CO2 and EOR Physics **Petrel Integration. **Optimization and FDP available. What other technologies you would suggest? See My Previous Post on Advanced Production Modeling Tools: https://shorturl.at/66uWy

Here's a nugget of today ;) What is Least Squares Migration in Image Domain in brief? It is a method designed to recover the true reflectivity of the Earth's subsurface by minimizing the difference between the imaged seismic data and the forward-modeled data. The technique involves using blurring operators, also known as propagators, and point spread functions (PSFs) to model the acquisition and subsurface effects. Objective of Least Squares Migration in Image Domain: Goal: To obtain a more accurate reflectivity model by minimizing the misfit between the observed seismic image and the modeled image. Misfit Minimization: This involves iteratively refining the reflectivity model to reduce the difference between the observed and synthetic seismic data. Forward-Modeled Data: Blurring Operators (Propagators): These operators simulate the effects of the seismic acquisition system and subsurface properties on the seismic data. They account for how the reflectivity is "blurred" during the imaging process. Regular/Irregular Diffractor Points: The forward-modeling process uses diffractor points to create synthetic seismic data that can be compared with the observed data. Point Spread Functions (PSFs): PSFs: These functions approximate the blurring operators and represent how a point source (or diffractor) spreads out in the seismic image, it is like a Green's function to describe the subsurface response to the propagated wavelet. Mathematical Representation: The mathematical representation of Least Squares Migration in Image Domain involves the relationship between the image, the blurring operator, and the reflectivity model: I=Hr where: I is the seismic image. H is the blurring operator (which are approximated by PSFs). r is the reflectivity model. The blurring operator H can be further decomposed as: H=M∗M' where: M is the modeling operator. M' is the migration operator. The result of the inversion process is a high-resolution reflectivity model that accurately represents the subsurface.

Data at your fingertips - #SCALDigitalization #AnalogSCAL Muhammad Nur Ali Akbar, our senior reservoir engineer expert, will present our groundbreaking innovation in native cloud solutions titled "Digitalization of Special Core Analysis: A Novel Solution for Constructing Synthetic Relative Permeability from SCAL Analog" at the Dig X Subsurface 2024 event organized by GeoPublishing Events. Join us at Dig X Subsurface! 📆 5-6 March 20224 🎯 Scandic Oslo Airport Registration: https://lnkd.in/dZeHpV99 #Abstract: This research delves into the evolution of utilizing Special Core Analysis (SCAL) data within the context of subsurface digitalization. The central focus is on the development of a cloud-native solution tailored to harness SCAL data from the Norwegian Continental Shelf (NCS) and optimize these resources for generating synthetic relative permeability curves without necessitating new core samples. This pioneering solution involves the digitization, structuring, and integration of SCAL data into a unified cloud platform, employing AI-based similarity ranking and SCAL trend analysis. The methodology seamlessly incorporates data collection, interpretation, and storage within the native cloud-based system, utilizing advanced statistical techniques to identify analogous SCAL data. It enables the anticipation of representative relative permeability curves, exemplified through field illustrations and uncertainty ranges based on the LET method. Key findings underscore three primary solutions: Core Data Management, SCAL Analog, and SCAL Trend Analysis. These solutions augment reservoir characterization, facilitate informed decision-making based on data, and provide support for comprehensive field applications. The platform's adaptability, transparency, and openness to further enhancements, such as SCAL upscaling and rock typing, hold promising potential for broader applications within the oil and gas industry. #SCAL #digitalization #transformation #coreanalysis #petrophysics #reservoirengineering #oilandgas #petroleumengineering #cloudcomputing #cloudsolutions #subsurface #database #cloud #nativecloud #datascience #dataanalytics #artificialintelligence #AI #machinelearning #AnalogData #conference

As professional data scientists in the oil & gas industry, we continuously seek innovative ways to enhance our operations and make data-driven decisions. One of the most groundbreaking advancements we're seeing is the application of Machine Learning to generate synthetic data. This technology is not just a game changer, it's revolutionizing how we predict and manage operational outcomes. Synthetic data, generated through sophisticated ML algorithms, allows us to create numerous scenarios and simulations without the need for extensive real-world testing. This approach is particularly useful in environments where data can be costly or hard to obtain. I'm thrilled to share our latest contribution to this field in a recent publication. In this paper, we explored how synthetic data can predict temperature logs, providing insights that are crucial for effective production engineering. Check out our findings here: https://lnkd.in/d7i8eF85 I invite you to delve deeper into our study, presented at the ADIPEC Conference, where we showcase the methodology and outcomes of this innovative approach. Discover how synthetic data and ML are transforming the oil and gas industry, driving efficiency, and fostering data-driven decision-making. #MachineLearning #SyntheticData #OilAndGas #Innovation #ReservoirManagement #SPE #ResearchAndDevelopment #TechnologyAdvancement

Importance of Geometric Correction for Data processing 1. How are CDP gathers made? CDP (Common Depth Point) gathers are a critical part of seismic data processing. A CDP gather refers to a collection of seismic signals (traces) recorded at different source and receiver positions, corresponding to the same subsurface depth point The creation of a CDP gather involves: Seismic signals are recorded from each source-receiver combination The signals corresponding to the same subsurface depth point from different receivers at various angles and distances are arranged into a CDP gather. 2. Corrections and stacking Seismic data is subject to various types of errors, such as location-based, time-based, and weathering-related errors. Weathering Correction: This accounts for the effect of the weathering layer that slows down the seismic waves. The data is adjusted to minimize this impact. Static Correction: Static correction adjusts for differences in elevation or position of the receiver or source, which can affect the travel time of seismic signals. NMO Correction (Normal Moveout Correction): The travel time of seismic signals varies due to the distance between the seismic source and the receiver. After applying these corrections, the traces are summed together in a process called stacking. 3.Stacking charts Horizons (Layers): The different layers or boundaries between them can be identified, which provide insights into the geological structures. Reflection Events: Reflection events in the stacking chart indicate the presence of subsurface structures, such as high-density rock formations. Faults and Deformations: The charts may also show deformations or displacements, such as faults or folds, indicating tectonic activity. 4. weathering correction Determining the thickness and velocity of the weathering layer. Adjusting the recorded seismic time to account for the slowing of waves through this layer. Adding or subtracting correction times to the recorded signals to remove the weathering effects. 5. Elevation correction Elevation correction is applied to account for the differences in height between the source and receiver locations. Obtaining precise location and elevation data for each receiver and source. Correcting the time shifts in the signals to place them on a uniform elevation basis. 6. Static correction Time differences in the signals are calculated based on the elevation and geographic position of the receiver and source. A fixed time adjustment is made for each receiver and source so that all signals are referenced to a common base level. 7.NMO correction: NMO (Normal Moveout) correction adjusts for the time delay caused by the distance between the source and the receiver in seismic data. The process involves: Calculating a correction time based on the source-receiver offset and the velocity of the seismic signal Applying this correction time to each trace so that all traces align at the same reflection event

𝗦𝗲𝗶𝘀𝗺𝗶𝗰 𝗗𝗮𝘁𝗮 𝗙𝗶𝗹𝘁𝗲𝗿𝗶𝗻𝗴 𝗧𝗲𝗰𝗵𝗻𝗶𝗾𝘂𝗲𝘀 1. **Bandpass Filtering:** Bandpass filtering selectively removes frequency components outside a specified range while retaining frequencies of interest. This technique is commonly used to suppress low-frequency noise (such as ground roll) and high-frequency noise (such as air blasts) while preserving seismic reflections. 2. **Denoising Algorithms:** Denoising algorithms, including median filtering, wavelet denoising, and empirical mode decomposition, aim to attenuate random and coherent noise present in seismic data. These algorithms exploit the statistical properties of noise and signal to enhance signal-to-noise ratio (SNR) and improve data quality. 3. **Spectral Whitening:** Spectral whitening techniques aim to flatten the seismic spectrum by equalizing the amplitude of frequency components across the seismic bandwidth. This approach enhances the resolution of seismic data and improves the interpretation of subtle geological features. 4. **Deconvolution:** Deconvolution techniques, such as predictive deconvolution and wavelet deconvolution, aim to enhance seismic resolution by compensating for the effects of the seismic wavelet and source signature. Deconvolution removes reverberations and enhances the definition of reflectors, particularly in complex geological settings. 5. **Adaptive Filtering:** Adaptive filtering algorithms, such as the least-mean-squares (LMS) algorithm and the recursive least squares (RLS) algorithm, adaptively adjust filter coefficients based on the local characteristics of the seismic data. Adaptive filtering is particularly effective for attenuating coherent noise and enhancing signal fidelity. ### Significance in Subsurface Imaging: 1. **Improved Interpretation:** Filtered seismic data provide clearer and more coherent images of subsurface structures, facilitating accurate interpretation of geological features, stratigraphy, and fault systems. 2. **Enhanced Reservoir Characterization:** By suppressing noise and enhancing signal resolution, filtering techniques enable geoscientists to delineate reservoir boundaries, identify fluid contacts, and estimate reservoir properties with greater confidence. 3. **Reduction of Imaging Artifacts:** Filtering techniques mitigate common imaging artifacts, such as multiples, ground roll, and acquisition footprint, which can obscure seismic reflections and complicate interpretation efforts. Photo refrence, credit : https://lnkd.in/df2bnKQs

𝐌𝐮𝐥𝐭𝐢-𝐂𝐥𝐚𝐬𝐬 𝐌𝐚𝐜𝐡𝐢𝐧𝐞 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐂𝐥𝐚𝐬𝐬𝐢𝐟𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐨𝐟 𝐆𝐞𝐨𝐩𝐡𝐲𝐬𝐢𝐜𝐚𝐥 𝐋𝐨𝐠𝐬: 𝐀𝐝𝐝𝐫𝐞𝐬𝐬𝐢𝐧𝐠 𝐈𝐧𝐬𝐭𝐫𝐮𝐦𝐞𝐧𝐭 𝐕𝐚𝐫𝐢𝐚𝐛𝐢𝐥𝐢𝐭𝐲 𝐚𝐧𝐝 𝐑𝐞𝐠𝐢𝐨𝐧𝐚𝐥 𝐃𝐢𝐟𝐟𝐞𝐫𝐞𝐧𝐜𝐞𝐬 𝐰𝐢𝐭𝐡 𝐀𝐧𝐚𝐥𝐲𝐭𝐢𝐪𝐮𝐬 In geophysical logging, the range of values for geophysical logs such as Natural Gamma, Gamma-Gamma, Neutron, and Photo-electric formation factor can vary significantly depending on the area and location. This variability is influenced by factors such as geological formations, mineral content, and other subsurface characteristics. Additionally, the calibration of recording instruments is not standardized across different regions and equipment versions, leading to inconsistencies in the recorded data. To address these challenges, it is essential to train machine learning models that are specific to the different versions of equipment and the various areas where the data is collected. This requires a robust framework that can handle the diverse data inputs, ensure accurate calibration, and maintain version control of the models. There is a growing demand for software solutions like Analytiqus, which can assist petrophysicists and geophysical logging engineers in quickly training and deploying machine learning models tailored to their specific needs. Such 𝐬𝐨𝐟𝐭𝐰𝐚𝐫𝐞 𝐬𝐡𝐨𝐮𝐥𝐝 𝐨𝐟𝐟𝐞𝐫: 𝐃𝐚𝐭𝐚 𝐈𝐧𝐭𝐞𝐠𝐫𝐚𝐭𝐢𝐨𝐧 𝐚𝐧𝐝 𝐏𝐫𝐞𝐩𝐫𝐨𝐜𝐞𝐬𝐬𝐢𝐧𝐠: Tools to efficiently integrate and preprocess data from different logging instruments and areas, ensuring consistency and accuracy. 𝐌𝐚𝐜𝐡𝐢𝐧𝐞 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐌𝐨𝐝𝐞𝐥 𝐓𝐫𝐚𝐢𝐧𝐢𝐧𝐠: Capabilities to train machine learning models on the processed data, taking into account the unique characteristics of each instrument and location. 𝐂𝐚𝐥𝐢𝐛𝐫𝐚𝐭𝐢𝐨𝐧 𝐌𝐚𝐧𝐚𝐠𝐞𝐦𝐞𝐧𝐭: Features to manage and apply instrument-specific calibration factors, allowing for accurate model predictions. 𝐕𝐞𝐫𝐬𝐢𝐨𝐧 𝐂𝐨𝐧𝐭𝐫𝐨𝐥: A robust version control system to track changes in the models, ensuring that updates and modifications are systematically managed and documented. 𝐃𝐞𝐩𝐥𝐨𝐲𝐦𝐞𝐧𝐭 𝐚𝐧𝐝 𝐌𝐨𝐧𝐢𝐭𝐨𝐫𝐢𝐧𝐠: Easy deployment of trained models and real-time monitoring to ensure they perform as expected in the field. 𝐔𝐬𝐞𝐫-𝐅𝐫𝐢𝐞𝐧𝐝𝐥𝐲 𝐈𝐧𝐭𝐞𝐫𝐟𝐚𝐜𝐞: An intuitive interface that allows users to interact with the software, visualize data, and interpret model results effectively. Software like Analytiqus can significantly enhance the efficiency and accuracy of geophysical logging analysis, providing valuable insights and improving decision-making processes in the field. For More Information visit: www.aicrux.co.in or https://lnkd.in/dHQG6KFS